A different type of AI company

Most AI companies (frontier labs, custom developers, and consultancies alike) refuse to take a stand on values for fear of losing your business.

Their websites cite bold values like:

• “disruption”

• “business innovation”

• “AI readiness”

• “unlocking value”

across paragraphs and paragraphs of (usually AI-generated) text devoid of meaning.

Here are our claims:

We hate slop.

• We do not publish AI-generated text or AI-generated images.

• The text on this website has been written by hand by our founders. Likewise for our LinkedIn posts and blog posts, and most importantly our email communications.

• We will not build tools for clients to mass-generate “slop” content.

• We are generally opposed to the use of generative models for sales purposes except in very specific scenarios.

Don’t be evil.

We strictly do not work with the following industries:

• Fossil fuels

• Tobacco

• Weapons manufacturing

• Private prisons

• Mass surveillance

Models should have specific tasks.

• Employees have specific roles, responsibilities, and procedures. We should have the same standards for models.

• If you determine the task to complete in advance, the uncertainty of ROI is only contingent on a model’s success at that task.

• Once a task is determined, success can be easily measured and tracked.

• Custom general-purpose chatbots are generally useless in the long term.

Generally, LangChain-type graph setups > agents

• Tool calling is often useful, but rarely the sole solution to a problem.

• Splitting a large task into smaller components allows the use of smaller, cheaper models while also boosting overall consistency.

The human touch is important.

• Sales relationships thrive on interpersonal connections. The proliferation of GenAI (especially slop spam cold outreach) makes this even more important.

• Generative models tend to produce low-quality output in creative fields, like graphic design and writing.

• People can often tell when writing is AI-generated, and it usually comes across as low effort and disrespectful to the reader.

• AI without human oversight is dangerous and bad for business.

Most GenAI projects are bad ideas and doomed to fail.

• Don’t just trust us; trust MIT’s NANDA lab finding that 95% of 2025 GenAI pilots provided no ROI.

• About 50% of projects were sales automations, which lost money.

• The projects that provide real value were the ones that automated tedious back office and operations procedures.