Perplexity Search as Code Lets AI Models Write ...

Your new post is loading...

Scooped by Charles Tiayon

June 8, 9:22 AM

Scoop.it!

From opentools.ai - June 8, 12:32 AM

Charles Tiayon's insight:

"Perplexity Search as Code Lets AI Models Write Their Own Search Pipelines
June 8, 2026
Perplexity has launched Search as Code, a new architecture that lets AI models write their own search pipelines in Python instead of calling fixed APIs. The approach cuts token usage by 85% and outperforms OpenAI and Anthropic on research benchmarks.

Why Fixed Search APIs Are Breaking AI Agents
AI agents have a search problem. The standard loop — model writes a query, search API returns results, model reads them, model writes another query — was designed for humans, not for autonomous systems doing hundreds of rapid searches. Context windows get stuffed with junk because the filtering logic is locked inside the search engine. The model can tweak the query but cannot control how results are ranked, deduplicated, or filtered.

Perplexity's answer, announced on June 6, is Search as Code (SaC) — an architecture where AI models write their own search pipelines as Python code and execute them in a sandbox. As 1 put it: "Instead of calling a ready‑made search API, models in Perplexity's new Search as Code architecture write their own search workflows as Python code."

How Search as Code Works
The architecture has three layers, described in detail in Perplexity's research paper:

Model (Control Plane) The AI reasons about the task, decomposes it, and generates Python code that assembles a custom search pipeline. It decides strategy — not just which keywords to use, but how to combine retrieval, filtering, and ranking.
Compute Sandbox A secure execution environment with a persistent filesystem for cross‑turn state. Chosen over a REPL approach because explicit state handling via serialization improves reliability on long research trajectories.
Agentic Search SDK Breaks Perplexity's search backend into atomic primitives — retrieve, fanout, filter, dedupe, rerank, parse_field — that the generated code can freely combine. No retrieval operation is dispatched through function calling; everything is code‑driven.
The Numbers — 85% Fewer Tokens, 4x Competitor Accuracy
In a case study tracking down 200 high‑severity CVEs (2023‑2025) with vendor‑specific advisory formats, Perplexity reported that SaC achieved 100% accuracy while consuming 42,900 tokens — an 85% reduction from the 288,700 tokens used by Perplexity's own standard pipeline. Competing systems scored below 25% on the same task.

Across benchmarks, Perplexity claims SaC outperformed 1 on 4 out of 5 tests. The largest improvement came on the WANDR broad‑research benchmark, where SaC showed a 45% absolute gain over Perplexity's own baseline. On DSQA, the gain was 29%.

Code Is the New Interface Layer
Search as Code reflects a broader shift in how AI agents interact with infrastructure. As Perplexity's research team wrote: "Code is a powerful medium for orchestrating preexisting capabilities — it can also serve as a gap‑filler for capabilities that aren't present in the search stack or SDK."

Rather than the model being limited to the parameters a search API exposes, SaC gives it access to the same primitives a human engineer would use to build a search pipeline — then lets it compose them programmatically. As The Decoder noted, the emerging paradigm combines models for strategy with deterministic runtimes for batching and filtering, with search infrastructure becoming an I/O layer rather than a black box.

What This Means for Builders
Search as Code is rolling out now in Perplexity Computer, the company's agentic AI platform. For developers building AI agents that need to do complex research — gathering data across hundreds of sources, verifying facts, compiling reports — SaC represents a fundamentally different approach from the "one query, one response" model that dominates current search APIs.

The implications are significant. If models can write their own search pipelines, the bottleneck shifts from can the API return the right results to can the model design an effective search strategy That favors frontier models with strong reasoning and code‑generation capabilities — and puts pressure on search API providers to expose their internals as programmable primitives rather than black‑box endpoints. Perplexity's SDK‑based approach may become the template for how search infrastructure is packaged for the agentic era."
https://opentools.ai/news/perplexity-search-as-code-ai-models-write-search-pipelines
#metaglossia
#metaglossia_mundus

No comment yet.

Concordia engineering student takes home top prize at Délie ta langue ! French-language public speaking competition | News - Concordia University

"Concordia engineering student takes home top prize at Délie ta langue ! French-language public speaking competition

Wendy Mbog’s winning speech encouraged women to tune out the noise and define their own paths

Wendy Mbog, a Concordia software engineering student, won first place at the 2026 edition of Délie ta langue!, the annual French-language public speaking competition. The accomplishment earned her a $5,000 prize, awarded by Quebec’s Ministère de la langue française.

Mbog’s presentation centered on the French expression, “Les chiens aboient, la caravane passe,” (“the dogs bark, but the caravan moves on”), which she used as a lens to address the pressures placed on women. Her speech explored the contradictory expectations women often face, encouraging them to step away from external judgments and define their own paths.

Claiming space through language

For Mbog, French is not just for academics or competitions. As her first language, it is tied to her Cameroonian origins and has shaped her experiences at home, in school and in how she engages with the world. She says she is drawn to the language’s richness and precision, and she admires those who can use it to express complex ideas with clarity and simplicity.

For Mbog, participating in Délie ta langue became a way to take ownership of her relationship with the language.

“I had been wanting to take part in a public speaking competition for a long time, to step out of my comfort zone and experience the power of spoken words,” she says.

Preparation and support

Mbog’s journey to the stage was supported by a network across Concordia. Espace Franco and Réussir en français coordinated her participation, with support from the Département d’études françaises, which provides mentorship to students from all disciplines.

“You could really say the whole university was behind her,” says Sophie Mailloux, Espace Franco manager and Concordia’s representative on this year’s Délie ta langue! organizing committee. “It was a proud moment to see francophone students who chose Concordia finding success and confidence in French.”

Pascale Sicotte, dean of the Faculty of Arts and Science, has attended the competition every year since Concordia began participating in 2022.

“It’s inspiring to see our students excel and share their passion for French,” she says. “Year after year, the candidates raise the bar. Standing on that stage takes courage, and their dedication and skill are evident in every moment of their performance.”

Mbog was also mentored by Julien Perrier-Chartrand, professor in the Département d’études françaises, and Noah Labranche, BFA 20 (Theatre), an actor and public speaking coach. She says they helped her refine her delivery and stage presence, complementing her own extensive preparation.

Where words meet action

Reflecting on her approach, Mbog connects her experience on stage with her engineering studies.

“My mother often says that scientists are first and foremost good writers: ‘It is necessary to understand the subtleties of a problem before being able to solve it,’” she says. “I think this also applies to the way I built my speech.”

For Mourad Debabbi, dean of the Gina Cody School of Engineering and Computer Science, Mbog’s success reflects a broader form of student development.

“It’s remarkable to see our engineering students embrace opportunities that cultivate leadership, communication and critical thinking — skills that are essential to their success as professionals and members of our community,” Debabbi says.

“Initiatives like this one allow students to grow in ways the classroom alone cannot.”"

June 8, 2026 | By Meera Nehme

https://www.concordia.ca/news/stories/2026/06/08/concordia-engineering-student-takes-home-top-prize-at-delie-ta-langue-french-language-public-speaking-competition.html

#metaglossia

#metaglossia_mundus

From www.concordia.ca - June 8, 2026 6:00 PM

Scoop.it!