Monkeynoodle.Org

This was a 300 person conference put on by the Product Management Club at UC Berkeley’s MBA program. About half students, half professionals (heavy on UCB alumni). I heard of it through Rands Leadership Slack but most people that I talked with heard about it through Lenny’s PM List. I did not attend the late afternoon and closing sessions.

Keynote, Todd Yellin

Todd Yellin
Culture is aspirational. If it was achieved and descriptional, that’s too low of a bar. Size changes a culture. Critical mass makes bureaucracy, culture counteracts that.
Support people making Decisive and Informed choices.

• Informed captain model is experts in each area.
• Farm for dissent (that’s not just asking if there’s any objections)
• Allow for diverse voices (actually check in with all the people affected)
• Disagree and commit
  The emperor doesn’t reside at Netflix… had a very smart CEO in Reed Hastings who led, yay, sometimes that feels like micromanaging, ugh. But he would set strategy and allow informed captains to do their thing. “We’re going to streaming” doesn’t mean CEO spec’s all the agreements and UX decisions and stuff. “We’re going international” — somebody’s got to decide subtitles or dubbing or both and how to price and how to cache.
  Big gamble in 2013: release all the episodes at once! Ted Sarandos’ decision (head of content at the time) as an informed captain. Product and Comms and Marketing and Content and Legal in the room, each of those teams had done their own investigation and conversation, Ted made the final call.
  Now there’s signal from product… what if we had interactive content? Todd tried this one out, gaming (still might go), interactive tv (didn’t go), “personalization was the most profound change”. BHAG: What if the interface could show you the one thing you want to see? Of course that doesn’t work, even if it did consumers would push back because that’s just psychology. Then you mix in some privacy options and it gets tougher. It’s a monthly subscription service, “product is on its knees begging for renewal every month”. So teams need to be empowered to make decisions. Have the discipline to push decision making down. Don’t fall for culture of consensus, it’s unhealthy and it takes too long. Leader needs to know which issues to force and which issues to let go on; know thyself and watch for your tendencies.
  Farming for dissent requires multiple channels, everybody shares on different levels and media. It’s not enough to ask in an open meeting, lol. Todd ran an office hours with open agenda, some people wouldn’t talk or attend but they’d comment on the google doc. Todd would then push the PM to address those comments as well as the talk. If PM wants to go against a consensus, Todd would allow: win is double plus great, but loss against consensus is bad.
  Interview question: You and nine others in a meeting with equal seniority/knowledge/interest. Where do you rank on how vocal you are (1-10)? In five years? The five years number is what mattered and the ideal is 5 or 6; a 9 is taking all the oxygen and a 1 isn’t adding the value they could.
  If you don’t lean forward enough to fall, you’re making a mistake of over-safety.

Q&A

Q as a PM how do you balance innovation and customer feedback?
A innovation for innovation’s sake is useless. Who’s the constituency? The customer, the investors, the content creators, the advertisers? Balancing their needs and conflicts will create innovation
Q How has being a PM help you become a director?
A That’s funny because I’ve been asking for funding and it’s a total “trust me, this is relevant experience” moment. Making a movie is a massively expensive ticking clock, and being really clear about what decisions need to be made and who will make them matters. That’s PM.
Q As a PM should you try to influence culture?
A As any individual you should! It’s half your waking hours, make it somewhere you want to be. I stayed at NetFlix 17 years because it was a really good match. Sure, if you need a paycheck then take it, but if you’re unhappy and trying to change it doesn’t work, GTFO.
Q how much did metrics matter to the model?
A I left before ads, so the core metrics we cared about were subscription retentions. 100k people in a test cell and 100k in a control cell, look for statistically significant quits first, then usage metrics. After those, signups were also interesting.
Q How did you filter signal from noise?
A It’s always about crossing the streams of passion and logic. Inspiring the people you work with to focus on what you’re trying to do, how to do it, what it makes everyone feel.
Q How do you keep teams from fighting each other over new ideas to do with AI?
A What kind of Darwinian culture is that coming from… Years ago, we started out streaming on the computer and were moving to TV, but we didn’t know what UX would work. So that’s a place where I did set up multiple teams, but they weren’t against each other, we weren’t going to fire people.
Q How do you balance farming for dissent with attaining consensus?
A To move slowly, focus on consent. It doesn’t slow things down needlessly to ask people what they think. To be clear, big decisions take a lot of time, and they should! We talked for years about going into sports, ads, gaming, they’d be hard to get back out of.
Q Why go into gaming? Just lucrative, or synergistic?
A Just lucrative is not a good reason to go into something, you have to have synergy with your strengths. We found that gaming spent huge amounts on marketing, as much as they do on making the game. With 300M househoulds, maybe 2B pairs of eyeballs, we had marketing and tie-in power that could work. Is it going to work? Unknow at this point. Was it a smart bet? Since it was my bet, I’m going to say yes, it was a brilliant move!

Fireside chat with Ami Vora and Ansu George

Ami Vora interviewed by Ansu George
Q you’ve talked about your non-linear career trajectory, from couch-surfing to PM. How did that happen?
A I was a PM at Microsoft and it wasn’t good for me, I bounced around in marketing and comms and stuff. So I eventually needed a job, and I really wanted to go to Facebook. I talked a friend into taking me around the office at Facebook and told everyone I met I wanted to work there. One person called me back for comms and explained all the reasons why they couldn’t hire me! But they could hire a temp, so I took a role at a temp agency and moved to the bay area. I felt a sense of magic around Facebook. People said things like “it’s more useful than my car, it tells me what my friends are doing and helps me understand the world.” I want roles that feel right.
Q I’ve also heard you say “build your career like a product.” How should the students here think about that?
A I try to use product principles in my life, like talking to my husband with extreme clarity. We invest so much in what we learn at school and work and then throw it out the window when it comes to careers. When you build a product you have hypotheses and then you pivot when they don’t work, that’s very normal. It’s really hard to accept that risk in taking a job! No one thinks worse of you for taking a job for a few months and finding it doesn’t work. It’s not irrevocable.
Q You got your job by taking any job and working into it. Times have changed, what’s different now?
A New tools have made it possible to build something faster, and so it’s easier to come into a cold outreach or opportunity ask with something. Put yourself in the shoes of “what someone would do if they had this job” and then take a couple of hours to do it. It helps people see what they’d get.
Q You also talk about upgrading yourself, what’s the next version of Ami look like?
A I wish I knew! We try to stretch ourselves but stay the same… what if we started over, honor the 1.0 but begin fresh with a 2.0? Every time I’ve had another kid, nothing we did before worked, it had to be started over. So what’s the next version? I’m really curious what it would be like to go back to taking bigger risks, maybe powered by these new tools. As a more senior person it’s harder to risk the established material.
Q How do you set goals?
A When I set goals for myself and my team, I’ve started to also share my personal goals, like “write some articles” or “take a two week vacation.” I’m a naturally awkward person, and I find it makes it easier for everyone to just embrace that, so put it into the goals. It models what the team should be able to do as well! It’s so low-risk. What’s the worst, no one’s going to say “don’t run that marathon”, and then people know things that you’re doing and it’s a little more comfortable. They’re research on how people, especially women, can be better at doing things when it’s on behalf of others. So I want to take a vacation: I model it as “i’m inspiring my team to be able to.” Little self-hacks that make it easier to operate yourself.
Q You were VP of WhatsApp but you took the notes in every meeting, what’s up with that?
A Some people think of note-taking as a junior role, but I think it’s important for executives to do services. Execs should be pouring drinks and cleaning up after a meeting. Also… note-taking is really a power position, history is written by the winners and what’s important is what the note taker captured. Sets the tone for what’s going to happen next. We all undervalue the power that we have, assume that we don’t have power, I was an outsider woman and felt I had no power in my early roles. Also people automatically don’t like people who have power, until you can show why you’re good for them. But taking notes, assigning action items, following up, it’s a low power appearing position that pivots into high power.
Q There’s not a lot of content about growing from Director of PM to VP to CPO, what’s that like?
A It’s kind of a trodden path to go from IC to director, different companies define it differently but there’s some clear things to do in each company. A lot of high performers expect there to be rules and paths to follow, and that doesn’t happen as you become senior. Your manager doesn’t know what you should do, they hired you to figure it out. You need to decide what to do and then get everyone to agree with that and help you out. “I think our analytics suck, so I’m going to work on fixing that, ask for budget and help, and run a monthly all hands to report on progress and it’s going to be risky.” As an IC you might hide your opinion but as a senior you have to say what you think and drive it forward. You can’t hide behind other people’s validation. But you also can get credit for the success.
Q How are you incentivizing your team to work together now that you’re a CPO? Goals? KPIs?
A It’s different in every team, you can talk to 20 different CPOs and get 20 answers. You have to fit into what the executive team needs, just like how every CEO does a different job. In my current role, lack of clarity was the problem, so I had to get every team to a one-sentence strategy and single metric that they could pursue. Then it’s possible to celebrate their successes. Any change has to be a process of doing it and then closing the loop with celebration because that’s how you train people to be successful.
Q Teams have mixed experience, what’s the right mix of people and competitiveness?
A My career has been one of unplanned success, I have friends that have long-term plans and then they do it, and I just don’t understand. I’m a bit of a dilettante, I worked in lots of roles and lots of products all around Facebook. I would do different things every time I’d get bored. It held me back from promotion, but it was what made sense for me, and later it worked out. It’s not even clear what the finish line is? But people who keep learning are the ones that succeed, and you can’t predict what thing you’re going to need to learn.
Q Looking at one of the posts on amivora.substack.com, you graphed frustration against opportunities and “what i feel ready for”, can you talk about those gaps?
A If you want to be president, you’ve got to wait until an election cycle. The timing might not be right. When you’re junior in a growing company you can get promoted once a year! There will not always be a VP or C-level position open, and you feel like “I am ready, give me the job.” I’ve started thinking about work like a sports team, you rotate people in and out based on what the team needs. Competing against another team might change up who you want to put in the field. Sometimes it’s the off-season, and you need to hit the gym and get stronger. I was complaining to a boss about this and he said “1 it’s time when it’s time, and 2, use this time to do the things that you’d wish you had done when you get that job.” Preparing for the job, but also taking the personal time that you won’t be able to then.

Q&A

Q what advice for an IC who wants to grow but doesn’t like politics?
A Politics is a tricky word because people use it to describe whatever social tasks they don’t like doing. If you want to get everyone to do things, you have to convince people. So if you want that to not appear like “politics”, you need to work somewhere that you’re pre-aligned with the decision-making process. If the org is CEO-driven and you don’t like convincing the big boss in big meetings, you’re not going to like it there. No matter what, you’re not going to grow past being an IC if you don’t learn how to influence people to agree with you.
Q What is the future of AI tools for PM?
A PMs have always used all the tools available to make better decisions on behalf of the customer. I grew up before server-side A/B testing, doing PM work at Visual Studio when releasing meant RTM to the CD printing factory. Using all the new tools to find the most efficient way to solve problems is the thing. There were a lot fewer PMs per company in the past, maybe that’s the future now? What we should do is to identify and channel the customer’s need. Maybe these tools mean fewer people doing all the other stuff.
Q How do you bring clarity when you know what you’re doing now isn’t right but you don’t know what to do next?
A Something I’ve found very freeing is to write out my fears. We talk about goal-setting but that’s really hard, you can get stuck in analysis paralysis: is this goal too ambitious, that goal not big enough. But fears are easy to write down. If I stay, what’s the worst that could happen, if I quit, what’s the worst that could happen. That helps me derisk. The worst thing ever is being in a job where you’re not 100% committed, you’re spending your energy spiraling on why you’re even here. So in that case, you can go run a job search. I’ll spend 12 weeks looking for a better role, and if I can’t find one, then I recommit to this role. That commitment stops the nonsensical spiral.
Q You’ve traveled around the world, how many countries?
A Three times! I don’t know, 25-30?
Q Favorite?
A I took a train trip across America, you can spend 24-72 hours in little towns that don’t have airports, I learned a lot about this amazing country
Q What technology changes your travel experience?
A cell phones are a life of luxury, I used to have to get local change and find a payphone to let my parents know I was okay. Guidebooks that might be out of date, that road’s not there! Having the Internet in your pocket for translation, payments, maps, it’s huge.
Q Any advice to students?
A you have very little control over the big things, you can only optimize for feeling lucky. Stay learning, keep looking for opportunities, ignore techs and titles and scan for luck.

Ajit Ghuman Art of SaaS, B2B Pricing Breakout

Ajit Ghuman
The Art of SaaS pricing

• hardware prices have high up front manufacturing cost and pricing is a math activity
• software prices have very low up front, and price is an art
• software as a service is somewhere between those
  Impact of pricing change on ARR (annual recurring revenue) and NDR (net dollar retention) needs to be measured and reviewed, pricing doesn’t get optimized often enough, money left on the table is the rule. Price is the easiest thing to change but it’s not strategic to alter blindly. If you are useful then customers will be happy to pay you. If your pricing metric is poor or your packaging is bad, they churn. Price is about alignment of the business to customer’s success. A lot of MBA training for pricing is based on CPG or B2C stuff, totally irrelevant for a B2B SaaS. We have packages, addons, metrics, and relatively small markets. For a B2C pricing is the only lever and statistical methods are more important for visibility and decision making.
  Who’s involved in the pricing decision:
• Sales <- -> Product Marketing via Sales enablement
• Rev Ops <- -> Product Marketing via Systems
• Pricing Team <- -> Product Marketing and Product via Pricebook
• FP&A <- -> Product Marketing and Product via Cost and Margin analysis
• Engineering <- -> Product via metering and features
• Product <- -> Product Marketing via GTM
  There’s two kinds of PMs: Product Managers are rare, Feature Managers are all over the place, disempowered clerks who just keep the wheels turning. Pricing isn’t relevant to Feature Managers so they don’t act on the strategy that the business needs. If PM doesn’t do pricing strategy, Product Marketing is the next on the hook. And they don’t either, Sales will. That’s real bad, because Sales will just do what’s right for the currently active deals without looking at the long term.
  5-step pricing transformation:
• Goals and segments. There’s no one perfect model. Your ICP probably isn’t clear. Get to good enough and refine later. What’s your margin goal? Market share goal? Cost reduction story? What is the nature and need of the customer segment(s)?
• Packaging. Tailor your package to the segments. BMW models for instance.
• Price Metric: Select a well-aligned pricing metric for your segment. You can change this easily.
• Rate-setting: Discount bands. You can change this easily.
• Operationalization: Measure and bill.
  The Geoffrey Moore Chasm slide, with discussion of how your pricing goals change over time. Early market, you need competitive pricing structures that lower client risk, well-designed packages. Crossing the chasm, you’re looking for Revenue predictability. In the main market it’s about sales efficiency, and at the tail you’re getting commoditized so you want defensive packaging and new innovation. “There’s a yo-yo effect of complexity to simplification to complexity, the reason for that maps to these changing company goals.”
  Most common packaging mistake for startups: Good/Better/Best (GBB) packages. It’s bad because it encourages discussing features instead of value, and introduces needless complexity. In B2B customers don’t fit your GBB packages. If they need one feature, they get discounted into a higher package and the rest of it is shelfware — and upsell blockers. Blind packaging is dumb.
  Packaging option spectrum: One Size Fits All (Netflix), Bundles + Addons (Slack), À la Carte (Medallia c 2016). Could be represented on a Flexibiity and Scoping complexity quadrant. A B2C Startup is probably closer to One Size Fits All, B2B Enterprise is probably closer to À la Carte. Products don’t change that much across the scale, they just get commoditized — which drives a need to reprice in order to keep up. Medallia’s margin got wiped out by Qualtrix as the market commoditized, that product wasn’t worth as much money any more.
  Packaging 5-step:
• Segment fit – are packages well tailored for your market segments? To do segment fit, list your features and t-shirt map them to how well they fit the ICP in a segment (fit here is “how important to the customer”).
• Upside capture – can you capture most of the revenue opportunity?
• Differentiation between packages – why does your customer pick one or the other?
• Scalability and growth – can you upsell and cross-sell?
• Recognize existing vs new customers – existing customer is easier to monetize than new one
  Discussion on publishing the price table on the website, enterprise typically doesn’t, PLG does, but a lot of PLG companies make the mistake of chasing the incorrect ICP though. Getting champions instead of buyers. It’s also really hard to have PLG + Enterprise together, then you get lots of cannibalization and in-fighting. Story about a company with large volume of PLG making little money and small volume of enterprise making all the money, PLG team got cut in the end.
  Pricing Metric Selection: Decision spectrum
• Capability pricing is fixed or predictable, vs Consumption is variable
• On premises is more fixed, single-tenant cloud is more fixed
• Multi-tenant cloud or a SaaS like S3 or OpenAI is more variable
  7 key considerations to a pricing metric choice:
• Customer’s risk perception – risky bets get lower prices
• Mental anchors – are they used to buying in a certain way? That limits your options.
• Alignment with value – do winning and consumption go hand in hand?
• Consumption pattern – does client consumption scaling impact your revenue?
• Cost patterns – does client consumption scaling impact your COGS?
• Competitive action – what is your competition doing?
• Implementability – can you accurately measure and predict this metric?
  Consumption and cost patterns metaphor, like buying a car (capability) versus using ride hailing service (consumption). Usage based pricing is faddish right now, but do your customers really want to use the service regularly or do they just want to have a capability? A car in the driveway is a predictable sale regardless of whether the driver uses it or not. Uber doesn’t make money if people don’t take rides. Two case studies:
• Kustomer: “We quickly learned, customers didn’t want innovative pricing. They wanted repeatable, consistent pricing that mapped to the budget they already had in place.” That’s an anchored market.
• Helpshift: Deciding what metric to use for a game support product. Selling to a publisher with a support team of 20 agents at $100 per month would be $24K ARR. 20 million MAU and $100/100K MAU, that’s ASP of $240K. Customer doesn’t want to buy it at $12K an agent, because that consumption is not aligned with winning. So don’t price by agent.
  Cost patterns: To maintain gross margin you may have to select a specific type of metric. Infrastructure SaaS has linear cost and value, that’s a better fit for consumption pricing (usage). Application SaaS, the cost shelves off while value keeps going exponentially up, that’s a better fit for platform pricing (capability). Ajit thinks AI is going to drive models back towards infrastructure instead of applications because every user action gets tied to an expensive compute-intensive task.
  Zendesk came out with outcome based pricing recently: you pay for cases resolved. Great idea to tie to customer value, but this metric is problematic: 1 define “resolution” 2 different outcomes are different sizes.
  Price Points: Veblen Effects are Real. A Rolex is a Veblen good, you can sell it to a high end market. A Swatch is a normal good, you sell it to a low end market. The markets don’t cross that line. Selling at the wrong order of magnitude just communicates that you don’t understand your market.
  Packaging and Price Metric hypotheses are informed by your goals and segments, then you do competitive comparison and get customer and market feedback. Testing can be done with primary research, Conjoint, Van Westerdorp. But in B2B, always go for primary research first.
  Common issues in Early Stages:
• they don’t have PMF, but try GBB packages that don’t correlate to real market segments
• they give away too much value in Free or Freemium
• they don’t change pricing when they grow into new needs
• they forget to think about upsell motions
• their packages are too tight or too loose
• they don’t monetize professional services
  Takeaways:
• Undertand the journey/stage of your company
• Start with customer segments and ICP
• Structure is way more important than Price Point
  Course: The Art of SaaS Pricing and Monetization by Ajit Ghuman and Jan Pasternak

Panel: Navigating the AI Revolution in Product Management

Panelists: Navnith Ramkrishnan, Chantal Cox, Neha Ruikar
Moderator: Raghu Ausoori
Q How has PM changed with AI?
Navnith: adds productivity, find new ideas and refine them. Get to prototype of a better customer journey quickly.
Chantal: we’re all going to be AI PMs because we’re starting to use the tools to enhance productivity. Some will be AI-enhanced PMs by continuing to focus on customer needs and understanding them, then using AI tools. Don’t be tech first. More junior PMs will be able to make better decisions sooner in their career because they’re going to have better access to data. No more write a PRD and wait two weeks for review.
Neha: AI won’t replace PM, but PMs that don’t use it could be. Use it to define your strategy, to do execution and run your metrics, to sync with your partners and keep everyone up to date. Gemini as a meeting secretary. AI isn’t creative, PMs are, so you be creative and use the tools for productivity. “AI is changing PMs from Excel and PowerPoint wizards to prompt engineers.”
Q How does AI improve your job as a PM?
Chantal: the tech isn’t new, we were using this at Amazon to promote products based on behavior, you see it all over. GenAI is not predictable, you get different answers all the time, but you can hackathon ideas really quickly. LTK quickly saw how to use it to auto-caption images, great idea, but getting to production quality is a killer. Only a couple of ideas make it to there. GenAI is also very expensive, is the ROI actually there? I don’t always see how to tie the goal to the customer’s action at a price they’ll pay.
Navnith: When I started at Microsoft we were using models to predict what customers would buy and recommend connections between people. At Tanium I’m doing endpoint management, developing trust in the system is even more critical. Think of self-driving cars, going from cruise control to adaptive control, lane-keeping, gradually increasing capability. So driving more workflows by granular improvement of workflows that customers want to have. The cost part is interesting, enterprises haven’t figured it out but the models should get cheaper. That judgement of when to use the tool to solve a problem is on the PM.
Neha: Yeah, cost is a problem. When I was at Twitter we used AI for personalization, and we tried doing things like moderation and brand safety with it, it was not great. At Paxos we can’t turn to a black box system for an answer, so we have to keep humans in the loops. AI to show you an ad is fine, no one cares much if it’s wrong. Approving your loan is a much different story. Another FinTech story is Web3 intersection, a lot more applications of AI there should help to keep things decentralized and private. Federated learning versus on-device learning are interesting. Lots of problems to solve to get to an AI-only product future.
Chantal: Ask ChatGPT for an itinerary, do you trust it blindly? If you ask it for a wedding dress and it shows you one, are you going to buy it? We’re complex people, we have trust issues, and we’re not going to just let the machine tell us what to do and click yes to everything. I see that itinerary suggestion, I think it’s going to be crowded because everyone else is getting the same recommendation and going there. The main skill of a PM is that you understand human behavior to the core. Your user probably doesn’t want to be one of hundreds of thousands. They’ve got their own pros and cons to address in any real problem.
Q AI potential for security problems, can you elaborate?
Neha: AI is being used for efficiency and automation in FinTech, but always with a human in the loop to ensure accuracy. That’s a guard rail that is critical, so it’s AI as enabler rather than replacement.
Chantal: Shopping is different, we’re going to optimize for the partner that pays bigger commission first, then the partner who sells sizes that fit you. We want to hit 80% on the customer’s need, remove friction where it’s the right thing to do… but wait, imagine a social network where people can post continuously, it’ll be junk. Sometimes putting a pause in is good. At CreditKarma, we had a signup process, making it super simple increased churn out of the process because users felt we were trivializing something that’s important to them. They had fear that their credit score would be hurt, for instance.
Navnith: Nobody wants to get their stuff patched, either. Making it super simple to do is riskier. AI gets better when it has more context. If I prompt it with what I want to do and what my constraints are and some things I don’t want, it’ll be better. In patching, we’ve had updates and rings and techniques for a long time, but customers need to be better informed how it can help them and how it can be used safely.
Q How to become an AI PM, how to use AI to get there?
Neha: look for opportunities and volunteer to do them. I learn best by doing, so I raised my hand to work on these problems. Companies care about profitability in this market, so try to position it as a cost saving. It’s a tough market to land a new job right now, so if you can’t do it in your current role consider hobby work.
Chantal: Don’t be afraid to try the existing tools. I started following LinkedIn stuff to get courses and ideas, asking Gemini for definitions. In the SF Bay Area we have excellent events as well, so I go to them. And of course, volunteer to work on it in your company. I think that this is taking down the barriers with other roles, I’m now sketching what I want in Figma and having Claude do a version. I used AI to set up A/B tests. I’m a director, I have reports, but I do IC work. You’re not expected to be a coder or a designer but you can meet them on their levels and use their tools. I have a strong opinion loosely held that PMs have gotten lazy, the DNA needs to be entrepreneur. PMs that are TPMs or fancy EAs are not meeting the potential of the role. I think that’ll help the ratio of successful products, which right now is not great.
Navnith: I don’t think PMs are actually lazy, but blocked. When I started at Microsoft I spent time learning KQL so I could use the data, now PowerBI is there to help so people are unblocked for data analysis. We improved the speed of development at another company by replacing the PRD review and approval process with an iterative ChatGPT style process. Spar with the bot, get to a decent level of detail and then try it out with the team. When building a product, the most important thing is the customer’s need, why would they want to do this, what will they get from it.
Q Understanding customer context and personalizing versus decentralization and privacy?
Neha: Consent Management Platform is the way, a lot of users, 60% are willing to share data to get a personalized result, need to make it so they can opt-out or -in and understand what their data will be used for. It’s hard to make a one-size-fits-all, though.
Q Humanizing context is more valuable, but overuse of tools reduces the humanity… how do you keep your environment from becoming slop no one wants to read?
Chantal: very important, can use the existing ranking and likes systems for that, content will die down if it isn’t creative and interesting. ChatGPT has a pile of documents, not the creativity of people. We will probably need to use agents to find and deprioritize bad stuff? If you spend a week posting ChatGPT stuff on LinkedIn you’ll probably tank your brand on your own. We don’t like to connect with a robot, we like to hear from real people with kids or pets. That doesn’t mean that we can’t have an AI agent do work on the other person’s behalf though, like an EA handling detailed follow up.
Q Tanium takes care of important stuff, how do you guardrail keeping employees from doing poor things with AI?
Navnith: We’re a Microsoft shop, we use their guardrails to make sure that the right levels of tagging and watermarking and content protection are in place. Customer data is kept far away from these systems.