About Heme

Companies should get better as they grow. Not more complicated.

We believe that AI should make businesses easier to run. Not give them more tools, more processes, and more work to manage.


That's why we're building Heme: to help companies turn their knowledge and ways of working into software that makes them more efficient, more capable, and easier to scale.

The Team

The team behind Heme

Professional Bussiness Profile Image

Carl Ottosson

Co-founder & CEO

Industrial Eng. & Mgmt, LTH

Professional Bussiness Profile Image

William Tanski

Co-founder

CS, LTH

Professional Bussiness Profile Image

Linus Settergren

Co-founder

Industrial Eng. & Mgmt, LTH

Professional Bussiness Profile Image

Axel Nilsson

Co-founder

Mechanical Eng., LTH

Professional Bussiness Profile Image

Philip Isakson

Co-founder

CS, LTH

Professional Bussiness Profile Image

Jakob Söderberg

Co-founder

CS, LTH

While helping companies implement AI, we kept running into the same problem: the models were getting smarter, but they still knew almost nothing about how the businesses actually worked.

We often had to stitch together company knowledge, workflows, and multiple AI tools just to get useful work done. Companies were adopting AI to make their operations simpler, but often ended up making them more complicated.

That frustration became the starting point for Heme.

What makes Heme different

We believe the next generation of companies will work differently.

One AI layer for the whole company

Company knowledge, personal Work Agents, custom apps, workflows and AI governance all live on the same foundation. Instead of stitching together separate tools for chat, automation, knowledge, app building and control, Heme gives the business one place to build and run AI. Unlike using ChatGPT, Lovable and automation tools side by side, Heme is designed as one operating layer for company AI.

The best intelligence per dollar

Heme isn’t tied to a single model provider. We can route work to the model that gives the best mix of quality, speed and cost, so customers benefit as the AI market improves instead of being locked into yesterday’s winner. ChatGPT is built around OpenAI’s models. Claude is built around Anthropic’s. Heme isn’t asking you to bet your company on one model provider.

One Work Agent for every employee

Every employee gets a persistent AI agent shaped by their role, knowledge, tools and authority. It knows the company context, but also learns what makes that employee’s work unique. ChatGPT and Claude begin as general-purpose AI assistants. Heme begins with the employee and the company they work inside.

Software that already knows the business

Apps and workflows built in Heme can inherit the same company knowledge, permissions and operating context as your Work Agents. You don’t have to explain the company again every time you build something new. Lovable starts with a prompt and builds an app. Heme starts with what it already knows about the company, then builds from there.

Enterprise-grade thinking. Mid-market focus.

Palantir connects AI, operational data, workflows, agents and governance through AIP, Foundry and its Ontology. Heme shares the belief that AI becomes valuable when it understands how the business operates — but we’re building specifically for 50–500 person companies that don’t want to deploy an enterprise data platform to get there.

One AI layer for the whole company

Company knowledge, personal Work Agents, custom apps, workflows and AI governance all live on the same foundation. Instead of stitching together separate tools for chat, automation, knowledge, app building and control, Heme gives the business one place to build and run AI. Unlike using ChatGPT, Lovable and automation tools side by side, Heme is designed as one operating layer for company AI.

The best intelligence per dollar

Heme isn’t tied to a single model provider. We can route work to the model that gives the best mix of quality, speed and cost, so customers benefit as the AI market improves instead of being locked into yesterday’s winner. ChatGPT is built around OpenAI’s models. Claude is built around Anthropic’s. Heme isn’t asking you to bet your company on one model provider.

One Work Agent for every employee

Every employee gets a persistent AI agent shaped by their role, knowledge, tools and authority. It knows the company context, but also learns what makes that employee’s work unique. ChatGPT and Claude begin as general-purpose AI assistants. Heme begins with the employee and the company they work inside.

Software that already knows the business

Apps and workflows built in Heme can inherit the same company knowledge, permissions and operating context as your Work Agents. You don’t have to explain the company again every time you build something new. Lovable starts with a prompt and builds an app. Heme starts with what it already knows about the company, then builds from there.

Enterprise-grade thinking. Mid-market focus.

Palantir connects AI, operational data, workflows, agents and governance through AIP, Foundry and its Ontology. Heme shares the belief that AI becomes valuable when it understands how the business operates — but we’re building specifically for 50–500 person companies that don’t want to deploy an enterprise data platform to get there.

One AI layer for the whole company

Company knowledge, personal Work Agents, custom apps, workflows and AI governance all live on the same foundation. Instead of stitching together separate tools for chat, automation, knowledge, app building and control, Heme gives the business one place to build and run AI. Unlike using ChatGPT, Lovable and automation tools side by side, Heme is designed as one operating layer for company AI.

The best intelligence per dollar

Heme isn’t tied to a single model provider. We can route work to the model that gives the best mix of quality, speed and cost, so customers benefit as the AI market improves instead of being locked into yesterday’s winner. ChatGPT is built around OpenAI’s models. Claude is built around Anthropic’s. Heme isn’t asking you to bet your company on one model provider.

One Work Agent for every employee

Every employee gets a persistent AI agent shaped by their role, knowledge, tools and authority. It knows the company context, but also learns what makes that employee’s work unique. ChatGPT and Claude begin as general-purpose AI assistants. Heme begins with the employee and the company they work inside.

Software that already knows the business

Apps and workflows built in Heme can inherit the same company knowledge, permissions and operating context as your Work Agents. You don’t have to explain the company again every time you build something new. Lovable starts with a prompt and builds an app. Heme starts with what it already knows about the company, then builds from there.

Enterprise-grade thinking. Mid-market focus.

Palantir connects AI, operational data, workflows, agents and governance through AIP, Foundry and its Ontology. Heme shares the belief that AI becomes valuable when it understands how the business operates — but we’re building specifically for 50–500 person companies that don’t want to deploy an enterprise data platform to get there.

Ready to grow without the growing pains?

Join the waitlist to be among the first to hear about Heme's launch and early-access opportunities (we won't spam you).

Welcome back, Alex

How can I help you today?

Prepare a proposal for our new client using our pricing, previous proposals, and the notes from yesterday's meeting.

Prepare a proposal for Nordica using the latest customer context.

Proposals

Meeting prep

Customer research

Pipeline

Prepare me for my next customer meeting

Review the Nordica opportunity

Find deals that need my attention

Extract action items and owners from this meeting note

Transparent image of sand dunes