Chris Bee

The hardest skill of being a B2B founder isn't building. It's telling the difference between someone being nice and someone being convinced.

18 Sept

Early on you take every polite "interesting, let me review with my team" as a signal. It isn't. Politeness is free and costs the other person nothing. Real signal shows up as an actual step forward, not a kind sentence.

18 Sept

Took me longer than I'd like to admit to stop grading meetings on how they felt and start grading them on what people actually did next.

Slack is repositioning hard as a cross functional AI hub, not just a chat app. That is Slack admitting the real prize isn't messaging, it's being the place AI reasons over what's happening in your company.

Problem is none of them want to be a neutral layer. They want to be the destination. So you end up with five different AI copilots that each see 20% of the picture and none of them talk to each other.

The company that wins here is not the one with the best AI inside its own app. It's the one willing to sit underneath all of them and actually connect the dots.

Every platform with a chunk of your company's data is going to make this same move. Notion will do it. Linear will do it. Your CRM will do it.

seattleaiweek.comSummit Speakers | #SeattleAIWeek

3 Sept

Excited to be speaking at Seattle AI Week this year. I’ll be talking about why I think the next big infrastructure layer is a shared knowledge graph. If you’re at the Summit, come say hi! www.seattleaiweek.com/seattleaiwe...

1 Sept

We put Claude with a stack of MCPs and research subagents up against Claude connected directly to Devplan. Same question. One had to find and piece together the context. The other started with it. Devplan returned a stronger answer, backed by evidence from across the workspace.

youtube.comDevplan MCP vs. Direct Connections in Claude: Faster, Cheaper, Evidence-Backed AnswersWhat happens when an AI assistant has to reconstruct product contex...

1 Sept

That is the value of shared product memory. Your agents should not have to rediscover the same context every time they get a new prompt. Watch the comparison 👇 www.youtube.com/watch?v=Qrg...

17 Aug

There's a moment I've watched play out at every company I've worked at. A team rolls out a new manual dashboard or a status tool, people use it for a few weeks, and then someone catches it being wrong or out of date. Nobody announces anything.

17 Aug

Everyone just quietly goes back to meetings, emails and DMs to fill the gap. That pattern shaped Weaver, the knowledge graph behind Devplan. It automatically ingests all relevant organizational data so every status report, dashboard or ad-hoc agent question has the latest understanding.

devplan.comInside Devplan’s Context Graph, WeaverBuilding a useful product knowledge graph requires more than storing data and running semantic search. Weaver is designed around trust: complete signal extraction, evidence-backed reasoning, no hallucinations, efficient retrieval, and a continuously updated understanding of your product and engineering organization.

17 Aug

My co-founder Anton wrote up how this works under the hood. It's the most complete picture we've published of our backend: www.devplan.com/blog/inside...

14 Aug

Spent the week talking to CTOs at mid-sized companies. Same pattern across all of them: AI spend approaching $1M this year, urgent focus on cost savings. We've quickly moved past the tokenmaxxing phase.

11 Aug

A question I keep getting: if Claude already connects to GitHub, Jira, and Slack, why would anyone need a knowledge graph?

11 Aug

Connecting AI to your tools solves the access problem. Understanding is a separate problem, and it's the expensive one. Every question pays the full cost of rebuilding your company's context from raw data, and that cost compounds as more of the company starts asking.

devplan.comWhy Companies Need a Knowledge GraphModern AI models excel at retrieving data, but they often struggle to turn that information into organizational intelligence. While direct connections to your systems provide access, they lack the ability to link decisions, context, and relationships across different platforms. A knowledge graph solves this by creating a shared memory layer. It continuously processes information to map how your organization actually works. This shift moves the effort from real-time reconstruction to pre-computed understanding. The result is better governance, lower costs, and the ability to capture intent rather than just outcomes. Information is becoming abundant, but understanding remains scarce. Knowledge graphs provide the context necessary to turn your data into actionable strategy.

11 Aug

A knowledge graph does that work once and reuses it everywhere. Full argument here: www.devplan.com/blog/why-co...

unite.aiWhy All Companies Need a Knowledge Graph in the AI EraAI has fundamentally accelerated software development, yet it has barely changed how software organizations operate. That mismatch is quietly becoming the biggest constraint on enterprise AI. For years, engineering teams...

4 Aug

Last week I published a piece on unite.ai about AI’s hidden coordination tax. AI is speeding up execution. Context is now the bottleneck. Read it here: www.unite.ai/context-eng...

29 July

Used to think the hard part here was getting AI to write good code reliably. Wrong about where the difficulty actually sits. The code part got solved faster than expected. What's genuinely hard is everything around it, the context before, the explanation after.

27 July

Every meeting cut for being inefficient was also a sensor. Killing the standup was usually the right call. Replacing what it was accidentally catching almost never happens. A lot of leaders have less visibility than a year ago and a faster team to show for it. Those two facts are highly connected.

23 July

Two people on the same leadership team, same project, completely different pictures of where it stands. Neither wrong exactly, just each working from a different slice of conversations over the past month. Happening in a lot of leadership meetings. Most never get to the out-loud comparison part.

21 July

Most bigger projects fail for the same reason: Somebody tries to lock in scope and timeline at the same time. That doesn't work. Pick one to hold firm and let the other move as reality changes, and the plan stays honest. Try to fix both at once and the plan doesn't bend, it breaks.

20 July

Connecting an AI assistant to your codebase feels like a real edge right now. Give it twelve months, every company will have done it. The companies that pull ahead build something above the connections, a layer that remembers instead of starting fresh every time.

17 July

Every time someone asks an AI assistant something spanning more than one system, it rebuilds the same understanding from scratch, then forgets it the moment the conversation ends. Ask a related question five minutes later, it does the whole thing again.

17 July

Paying once and reusing that understanding everywhere is a different economic model. Most companies are still on the expensive one.

16 July

People ask if @DevplanAI is like Notion or Linear, trying to find the neatest category to put us in. Fair question. Those tools assume you already know what to track. We sit one layer above that, automatically figuring out what's happening before anyone opens a tool to look.

15 July

Delegating to an agent rewards a skill most of us never developed: saying exactly what you mean. Closer to a fast new hire with zero institutional memory than someone you've worked with for years. Knowing exactly what you meant before you said it is turning out to be the actual differentiator.

14 July

Every company I talk to is generating more information than ever, more code, more threads. Almost none are generating more understanding at the same rate. Those used to track together. They've come apart completely, and that gap is the real constraint most companies haven't named yet.

9 July

Ask around what "Customer Onboarding" means at most companies. Product means the roadmap initiative, support means the weekly workflow. Both right, technically. Hand that term to a few AI agents working independently and each one builds its own confident, completely different idea of it.

8 July

Every org I've been part of burns real time on two things: tracking plan versus actual, and moving information from where it was written to where someone needs to read it. Neither is the work. It's the tax for not having a better way to keep everyone looking at the same picture.

6 July

The reasoning is buried in a Slack thread. The ticket is in Jira. The actual change is a GitHub commit. The plan someone wrote weeks ago is sitting in Notion. None of those four things talk to each other, and that's exactly why risk shows up late. Devplan pulls it all into one shared context.

1 July

Most project tracking is theater. Someone asks for a status update. Someone else spends 30 minutes assembling one. We think software should do that instead.

1 July

Devplan automatically tracks project progress from code, tickets, docs, Slack, and meetings, so everyone stays aligned without chasing updates.

29 June

Thanks to Nathan Owen and his write-up on @devplanAI and @grandvcp’s investment thesis. It's a sharp take on the “coordination tax” in product teams, and why a shared knowledge graph can unlock execution: nathanowen.substack.com/p/devplan-a...

89% of leaders say AI has sped up work. Only 6% can point to measurable company-wide results. This is AI fragmentation tax. It costs organizations $161 billion in lost productivity each year.

That drives up costs, slows down answers, and produces inconsistent results. Weaver works differently. It continuously processes organizational signals, so the AI already understands your product before anyone asks a question.

techfundingnews.comExclusive: Ex-Uber and Amazon alums raise $2.5M to reduce AI coordination costs by 3.5x — TFNDevplan, a Seattle startup founded by ex-Uber and Amazon executives, has raised $2.5M in seed funding to build a real-time knowledge graph that reduces AI coordination costs by 3.5x.

The result: queries run 2x faster and 3.5x cheaper than standard setups, and product managers are reclaiming 8 hours a week. Thank you to @Sofia C and @TechFundingNews for featuring our launch. Read the full story here: techfundingnews.com/devplan-2-5...

89% of leaders say AI has sped up work. Only 6% can point to measurable company-wide results. This is AI fragmentation tax. It costs organizations $161 billion in lost productivity each year.

The reason is simple. Most AI tools treat every query like a new research problem — scanning repos, tickets, and docs to reconstruct context from scratch.

24 June

Most leaders don't need more updates. They need help understanding what matters. We built Daily Digest to connect activity across your organization into actionable insights and recommendations.

pulse2.comDevplan Raises $2.5 Million Seed Round To Build Product Intelligence Platform For Software TeamsDevplan announced its launch and the closing of $2.5 million in seed funding. The round was led by AI2 Incubator and Acequia Capital, with participation from Mighty Capital, Grand Ventures, and eLab Ventures.

23 June

Grateful to see the response to what we've been building. The problem clearly resonates, which I suspected, since every product leader I've spoken with this year has described some version of it. We're just getting started. pulse2.com/devplan-rai...

luma.comWhat Serious Founders Should Build Now — with Stripe, ArchAstro, and Devplan · LumaWhere is AI truly delivering value in enterprise settings, how are team structures and roles changing, and where are the biggest opportunities still up for…

22 June

Startups used to win by moving faster than incumbents. AI was supposed to widen that gap. I'm not sure it's working out that way. Talking through it with Vivek Sharma and Rajeev Rajan at AI House Seattle. Real adoption, real teams, honest takes. luma.com/aihouse-x3rk

Today, we launch Devplan. We're also announcing $2.5M in seed funding led by AI2 Incubator, with participation from Acequia Capital, Mighty Capital, Grand Ventures, and eLab Ventures.

What we're seeing so far: • Product and engineering managers reclaiming 8-10 hours per week • Context queries running 2x faster and 3.5x cheaper than standard AI workflows connected directly to source systems • Great feedback in production with dozens of fast-growing teams

We believe every software organization will eventually have a system responsible for maintaining shared understanding. We are building it.

Today, we launch Devplan. We're also announcing $2.5M in seed funding led by AI2 Incubator, with participation from Acequia Capital, Mighty Capital, Grand Ventures, and eLab Ventures.

The bottleneck is no longer building software. It's maintaining shared understanding. Devplan connects tools teams already use into a shared product intelligence layer. Our knowledge graph, continuously understands what changed, why it matters, and what needs attention across the organization.

Product and Engineering leaders get real-time visibility into progress, risk, decisions, and opportunities. Teams spend less time searching for information, assembling status updates, and reconstructing context. AI agents gain access to the same understanding as the humans they work alongside.

Today, we launch Devplan. We're also announcing $2.5M in seed funding led by AI2 Incubator, with participation from Acequia Capital, Mighty Capital, Grand Ventures, and eLab Ventures.

But while software development accelerated, the systems used to coordinate product development barely changed. Teams still spend hours every week trying to answer simple questions. • What changed? • Why was this prioritized? • Which projects are at risk? • How does this feature work?

The answers exist. They're just scattered across Slack, Jira, GitHub, docs, meetings, and a hundred other places. As AI accelerates execution, this problem gets worse. More code. More fast decisions. More context. More information spread across more systems.

Today, we launch Devplan. We're also announcing $2.5M in seed funding led by AI2 Incubator, with participation from Acequia Capital, Mighty Capital, Grand Ventures, and eLab Ventures.

My co-founders and I started the company because we believe software development has fundamentally changed. AI has made writing code dramatically faster. Small teams can now build what once required entire organizations.