Substrate - summary

Shared by Pavel · Updated 2026-11-03

Substrate is building the memory and context layer that lets people and AI agents work from the same current, permissioned, human-readable company knowledge—and early customers are already paying for it.

Updated 1 Nov 2026 with new facts.

The problem: AI lacks dependable company context

AI agents can act only as well as the information they can access and trust. In a company, that information is spread across tools and people; it changes, carries different permissions, and often needs human interpretation. Without shared, current context, teams cannot confidently delegate meaningful work to AI.

This is an infrastructure problem, not a prompt-writing problem. Companies need a way to make knowledge usable by both people and agents while retaining human control over what is shared and changed.

The product: shared memory that people can inspect and control

Substrate is a permissioned, human-legible memory and context layer for people, teams, and their AI agents. It combines editable, Notion-like pages with scoped sharing, and supports connected information sources including Gmail and Notion. (more: MCP and connected context)

Its defining choices are deliberate: readable, permissioned, and owned. People can edit the memory, govern access, and keep ownership; agents can use up-to-date information to run workflows autonomously within those boundaries.

Substrate integrates through MCP and works with Claude and ChatGPT, placing its context layer alongside the AI tools teams already use.

The opportunity: become the context layer beneath AI work

As companies move from AI assistance toward delegated workflows, reliable shared context becomes a foundational requirement. Substrate is designed for that role: one governed layer that serves people, teams, and multiple agents rather than a separate, opaque memory for each interaction.

The product’s architecture addresses a core adoption barrier: companies need AI to be useful without surrendering control of their knowledge. Human editability, scoped permissions, and ownership make the memory legible and governable while still enabling automation.

Construction and renovation are Pavel’s intended starting market, where Substrate can focus on practical workflows and customer outreach before expanding its reach.

Early validation: customers are paying in open beta

In six weeks of open beta, Substrate reached 9 paying customers with zero churn and 41 real signups since September 11, 2026.

Those customers came through multiple early channels: five from a hackathon, two from LinkedIn content, and two from X and LinkedIn outreach. This is an initial, channel-tested foundation—not yet proof of a repeatable acquisition engine. (more: Customer traction: counts and early spend)

The traction establishes a meaningful starting point: users have encountered the product, converted to paid use, and remained customers through the reported beta period. The next commercial task is to test messaging at greater scale and turn early channel learning into a repeatable growth process.

Founder-market fit: engineering credibility and commercial execution

Pavel Tkachyk combines civil-engineering training with roughly a decade of experience at Hilti, including global product management and marketing leadership. At Hilti, he led 77% sales growth across nine countries, a record of commercial execution relevant to taking a technical product to market.

He is now building Substrate as founder and CEO while studying as an MIT Sloan Fellow. His background connects the company’s construction starting point with experience in product, marketing, and customer-facing growth.

The investment case

• A consequential layer in the AI stack: Substrate addresses the persistent context problem that sits between company knowledge and useful agent action.
• A product designed for trust: Human-readable memory, scoped permissions, editability, and ownership make the system governable as well as useful.
• Evidence of willingness to pay: Nine paying customers and zero churn during six weeks of open beta provide an early commercial signal.
• More than one early acquisition route: Customers arrived through a hackathon, LinkedIn content, and direct social outreach.
• Founder with relevant domain and go-to-market experience: Pavel combines civil-engineering background with a documented commercial growth result at Hilti.

The case is compelling because Substrate pairs a clear infrastructure thesis with a differentiated, governable product and early paying use. The opportunity now is to prove repeatability: sharpen the highest-value workflow, establish a focused initial market, and scale acquisition from the channels that are already producing customers.

What the next stage must prove

The next stage is to convert early validation into a durable growth engine: identify the workflows customers value most, test positioning across a larger audience, and show that customer acquisition can scale beyond founder-led early channels.

Substrate should be presented as an early company with meaningful initial validation—not as a scaled business. The investor proposition is that a permissioned, human-legible memory layer can become essential infrastructure for teams adopting AI agents, and Substrate is already testing that proposition with paying customers.