Introduction

After more than two years of research and experimenting, Convictional is reintroducing ourselves. Today we're launching the first product created to solve goal alignment in organizations for good. Our mission is to keep teams mission-aligned by design, building a healthier replacement for work chat tools. I'll share more details here on the problem we solve, how we are solving it, and what I've learned using it to date.

What we solve

Most of the software we use day to day at work was built for a world where people do task work themselves. As adoption of agents grows, our role as humans focuses less of our time and energy on tasks. We need to integrate AI into work in a way that deliberately and permanently preserves human taste and judgment. If we don't, we lose the ability to preserve knowledge, make decisions and pursue our goals in the long run.

Companies increasingly have a set of common and difficult organizational problems:

• Task work speeds up, while judgment work happens at the same rate. People spend more of their time searching for context to feed agents and answer questions in order to make decisions.
• More things happen because execution is cheap, but they end up being the wrong things and not leading to outcomes you want. The faster work goes, the more important goal alignment becomes.
• As agents generate huge volumes of content, that gets mixed with organizational knowledge and slop compounds rather than preserving the human tacit knowledge you need to win in the long run.

Convictional has observed these problems, in our prior business and in many conversations since. We have met with CEOs, Chiefs of Staff and BizOps leaders in companies ranging from 2 to more than 2,000 people. We have conducted hundreds of research interviews, and are working on publishing content to share our learning. The bottom line is the problem is real, and it's getting worse as work speeds up. Every organization is going to need a solution to this problem in order to remain relevant and differentiated.

How we solve it

Convictional is where high quality context in your organization lives. As you use us for team chat, as an inbox, for docs, in your meetings and to set and align to goals, the built-in company brain improves its knowledge of your organization. We then connect to the agents you are using today via MCP.

We have been working on research for this concept for over two years. We have posted elsewhere here about what we learned. A lot of companies are combining Slack, a company brain, Notion or Docs, Granola or similar, a CRM and agents together. This works for some kinds of tacit knowledge. The reality is though, without bringing understanding together in the place where the judgment work happens in close proximity, it doesn't work nearly as well as it could in one system. We have shared more writing on that here.

There are several ways our approach to solving this problem is unique:

• Taste and judgment work requires uninterrupted attention. We batch notifications, and filter them by organizational goals in a unified inbox that stays a calm, prioritized, singular list.
• Company goals are the organizing principle of the software that everything is organized around. Goals live in the app with rich context about updates and changes, and can be used to organize other work.
• Decisions and who made them are recorded explicitly by people. A little button found in chat, collaborative email, doc comments and posts can record decisions where and when they happen. When you do this, related context is gathered together and stored in the company brain.
• Autonomous agentic research agents can search for anything you need in the company brain in a way that is more token efficient and rich in citation, especially for hard to find answers with diffuse sources. We apply many fancy techniques to enable this. More of our research is here.
• Beyond that, we don't offer write-capable agents ourselves, preferring to expose the company brain context via MCP connection through our tools and plugins to connect with your existing agents.

Our approach is focused on the work that remains for people after AI: exercising taste and judgment, and doing it as well and as fast as we can. We are not focusing on task automation, since AI labs are already making fast progress there and increasingly the models behind the agents are becoming commodities. In the long run, organizations' ability to create new knowledge, make decisions well and align the work they do to their goals will be what endures. We depend on AI, but it also depends on us, and that is permanent.

What I've learned using it

We have been using variations of our own product internally for almost as long as we've been researching this problem. I've learned a lot about what works, and even more about what doesn't.

We often joke about how the product we're building has to 'hide the vegetables'. The reason being, you can already use AI as a socratic reasoning buddy in chat form to help avoid biases and make good decisions. It can be draining to stress test your decisions this way though, which is why most people most of the time don't do it. The product has to feel satisfying to use, while achieving better, faster decisions.

There are a lot of practices that we have published in our company handbook that could help any organization whether or not you use our software to get closer to this vision of judgment centered work.

Here are some of the changes we expect organizations will have to make as a result of agent adoption:

• Implement explicit goals: It doesn't matter if you use OKRs, EOS, SMART or otherwise. Goals have to be explicit from day one to align people and AI.
• Manage your prior job: AI is for task work, we're all becoming the managers of our prior job, focused on exercising creativity, taste and judgement. Focus development plans on levelling people up into managers of their prior role.
• Record decision processes: People and AI need shared context about decisions, and a way to search across them next time things come up.
• Share tacit knowledge: Publicly share discussions across the company to compound cross-functional knowledge and organizational memory.
• Formalize proposals: Have a place where people go to propose new ideas.
• Formalize process changes: The same process engineers use to make code changes should apply to content, SOPs, decisions, etc.
• Set roles by judgement domains: Job descriptions should be organized by decision type rather than task type (tasks are being automated).
• Goals as organizing principle: Decisions and projects should be organized around the explicit goals rather than disconnected from them
• Optimize culture for focus + connection: Balancing focus (deep work) and connection (collaboration) is essential. Too little of either is a serious problem.
• Writing is thinking: Protect the difference between AI and human writing. Humans write to think, AIs write to meet content obligations. Oil and water.
• Effectiveness not efficiency: Performative indicators that people are working are obsolete, all that matters is 'effect on goals', not presence

We have learned that adopting agentic AI in particular requires total reorganization of how we work with each other, how roles on the team are defined. It's a really challenging thing that most companies are only very early in taking on. As the models become more powerful, the urgency of goal alignment becomes more pronounced. The time finally felt right, with more people recognizing this and searching for solutions, to share both our product and our learning with the world. We're excited to play a role in fixing this.

If you want to talk about how these changes are impacting your organization and what you can do about it, I'm focusing my calendar on 1:1 demos and problem solving conversations over the coming weeks. If the problem resonates, we're inviting teams to sign up starting today: https://convictional.com/signup