# The 8 Levels of Agentic Engineering — Bassim Eledath

<https://www.bassimeledath.com/blog/levels-of-agentic-engineering>

- Host: [bassimeledath.com](https://links.thingelstad.xyz/hosts/bassimeledath.com/)
- Saved: March 11, 2026 (2026-03-11T21:16:10Z)
- Record: https://links.thingelstad.xyz/links/b971f377dd65/
- JSON: https://links.thingelstad.xyz/links/b971f377dd65.json

## What Jamie wrote

Great article that is framed on engineering but could really be any domain that has similar characteristics. The opening paragraph frames the question right.

> AI's coding ability is outpacing our ability to wield it effectively. That's why all the SWE-bench score maxxing isn't syncing with the productivity metrics engineering leadership actually cares about. When Anthropic's team ships a product like Cowork in 10 days and another team can't move past a broken POC using the same models, the difference is that one team has closed the gap between capability and practice and the other hasn't.

It is a pretty common comment to hear "we adopted AI coding tools and got slower". The pattern is pretty simple. As agents create more code, if the "human in the loop" insists on doing a detailed review, you are throwing away all the benefit you could have received. Then, since the AI agent can produce orders of magnitude more code than can be reviewed, you jam the system and output plummets. 

**This is why I've switched my focus from an efficiency mindset to a throughput one.** 

Enabling agentic capability is measured by enabling "machine speed" on an entire function.

The 8 Levels identified are good.

> 1. Tab Complete
> 2. Agent IDE
> 3. Context Engineering
> 4. Compounding Engineering
> 5. MCP & Skills
> 6. Harness Engineering
> 7. Background Agents
> 8. Autonomous Agent Teams

My main edit would be that these are not a progression. You can move forward in more than one at a time, but I would agree that you need to carefully consider dependencies and connections.

## Facts

- Kind: [Page](https://links.thingelstad.xyz/kinds/page/)

## Appears in

- [WT343: Weekly Thing 343 / Commune, Chaos, Renaissance](https://weekly.thingelstad.com/archive/343/), Notable

## Saved the same week

- [When Using AI Leads to “Brain Fry”](https://links.thingelstad.xyz/links/26437052b394/) (hbr.org, March 11, 2026)
- [Run prompts on a schedule - Claude Code Docs](https://links.thingelstad.xyz/links/6048b762e0db/) (code.claude.com, March 11, 2026)
- [OpenAI’s new GPT-5.4 model is a big step toward autonomous agents | The Verge](https://links.thingelstad.xyz/links/365cadf028a4/) (theverge.com, March 11, 2026)
- [The View From RSS](https://links.thingelstad.xyz/links/004fd870b7a3/) (carolinecrampton.com, March 11, 2026)
- [10 Must-Have Skills for Claude in 2026 | Medium](https://links.thingelstad.xyz/links/de787a3e1f9b/) (medium.com, March 11, 2026)

## Elsewhere

- Wayback Machine: https://web.archive.org/web/*/https://www.bassimeledath.com/blog/levels-of-agentic-engineering