How Tato gives its team and AI agents instant access to live company knowledge

Tato needed company knowledge that could keep up with the company

Tato moves quickly. The team works closely with customers, ships frequently, and uses AI agents throughout product development.

The knowledge behind that work was spread across code, Linear, Slack, Granola, and the founders themselves. Tato had built a product-intelligence skill for its agents, but it captured the product at one point in time. As the company shipped new features, that context fell out of date.

Tato connected those systems to Devplan. Devplan organizes their activity into a knowledge graph that stays current as the product, customer needs, and company decisions change.

That gives Tato one source of company context that both people and AI agents can use without reconstructing the history themselves.

Customers hear about relevant new features faster and more consistently

Before Devplan, Tato frequently shipped a fix or feature for one customer that would also benefit others, but did not always have time to identify and notify everyone for whom it was relevant.

Sahil wired Devplan's knowledge agent into a small internal skill that turns "what shipped this week" into customer-specific release drafts. Each week, Devplan connects product changes with the services each customer uses, then drafts personalized updates in Tato's voice. The team starts with a useful draft instead of reconstructing the release history by hand.

"The change summaries from you guys are honestly already customer-grade, so we just light-edit and don't have to rewrite. That's the whole reason it works."

A person still reviews every update. Devplan does the time-consuming part: matching changes to the right customers and drafting the message.

"Before you ask us, ask Devplan"

When Neha joined Tato in customer success, Ana and Sahil made Devplan part of her onboarding from the beginning.

"She was the first person we said, 'Before you ask us, ask Devplan and see what happens.'"

Neha uses Devplan for the questions that would otherwise interrupt a founder: how a staff user changes a membership, whether an existing capability can support a customer request, or what might cause a payment problem.

The founders stay focused on customers and product work, while Neha gets enough context to keep moving.

New hires use Devplan for the first pass and bring Ana or Sahil only the questions that require judgment or new context.

"There's no point in hiring if I'm just going to slow you down. So how can we solve for that?"

With less onboarding work routing back to Ana and Sahil, Tato can hire more aggressively without assuming that every addition will create weeks of founder interruptions.

Claude and Codex automatically use current company context

Tato is a deeply agent-driven company. Claude and Codex are part of how the team plans, builds, reviews, and improves its product.

Sahil added Devplan to Tato's CLAUDE.md and AGENTS.md files, the instruction files that guide Claude and Codex inside a repository.

Now Devplan is invoked automatically whenever a request is relevant to product work, planning, or implementation. When an agent reviews a ticket, plans a feature, audits the backlog, or evaluates an implementation, it can ground the work in Tato's current product, customer, and delivery context.

Because the instruction lives in the repo, the team does not need to remember to call Devplan. Claude and Codex pull it in while the work is happening.

One knowledge graph, used across Tato

Tato connected Devplan to the systems where its work already happens. That same knowledge now helps the team identify which customers should hear about each release, helps new hires answer questions, and gives Claude and Codex current context.

Faster hiring is one consequence. The broader result is that Tato's people and agents can use the company's knowledge without depending on a founder to find, explain, or update it.

← All posts