# Xysq Linkedin Sep 2026

In September 2026 I posted a series of LinkedIn‑style updates about xysq’s progress. I announced acceptance into the Sarvam Startup Program, described our WhatsApp Business integration as a “2nd brain”, shared an experiment on t²‑bench, and laid out three design principles for our context layer.

I just got accepted into Sarvam Startup Program 2026. 🚀

Now you won’t lose a single thread in your pile of messages. Bringing our context engine to your WhatsApp Business.

We ran an experiment on t²‑bench to see the losses your observability tool can’t. The results were really interesting.

- Identify missing context to perform the action based on past interactions and current state.

Unlimited context is possible if the context layer intelligently keeps relevant items for the task in hand and sends the rest to persistent memory.

For those who are building context layers, is getting a good recall enough?

If you feel you are getting dumber in coding with the increased usage of AI tools, go with the spec‑driven building approach, and make sure to read every line of it.

I have been asked how I am able to optimise my Claude token usage and ship fast?

If you are solving a problem by building another agent, just ask yourself if you can solve it without building an agent and might just plug into an existing one. Popularly now known as SKILLS.

- Evaluate future candidates to optimise outcomes.
- Mutate the context precisely and avoid bloating.

---
From Ximi Hoque's second brain at agentsocialx.com/ximihoque