Content moderation

Read the room, like our government clients

The world is changing faster than ever – and so are the stories people tell. From tracking emerging narratives and disinformation to flagging possible hate speech, the only way to stay ahead is through AI.

Content moderation is where Pienso began, and where we continue to lead. Our nuanced content flagging accounts for subtleties such as neologisms and slang. And our models can be adapted as fast as language changes.

Pienso identified COVID-19 antivirals before the CDC did
— and helped a government agency track disinformation.

  • Our work with government clients has enabled them to use and adapt Pienso for the vital and complicated business of tracking information as it evolves. At the outset of the Covid-19 pandemic, a US intelligence agency used Machine Learning to better understand SARS and Covid-like diseases. (Intelligence Client 1) Another member of the intelligence community uses Pienso to analyze media for sources of disinformation and to track narratives. (Intelligence Client 2)

Topic Classification

How do we find out
what people are saying?

Pienso classifies each conversation into topics, as defined by the people who know the subject best. While generic AIs rely on low-accuracy techniques, Pienso’s fine-tuned classification allows for nuanced content flagging – even over content that’s diverse, free-flowing and nonstop.

The solution

For Intelligence Client 1, Pienso immediately ingested all WHO research papers on SARS/Coronavirus up to that point. In just two weeks – before other vendors had even laid their groundwork – Pienso delivered a topic-trained model that identified the antivirals, including Remdesivir, that were likely to be effective against Covid-19. This helped the agency secure the supply of the drug.

Pienso’s platform immediately ingested the massive volume of text data comprising all WHO papers on SARS/Coronavirus to that point. Within 2 weeks – and while other vendors were still laying the groundwork for their anticipated months-long effort – Pienso delivered a model tailored to the agency’s needs. Before the CDCs realization of the same, Pienso’s model identified the antivirals (including Remdesivir) that were likely to be effective against Covid-19. The agency was able to take steps to secure the supply chain.

Gov1
Threat identification

How do we know
what they really mean?

Social media posts and news articles can provide advanced warning of potential violent acts. But the nuances of social media, tone, and sarcasm make automated recognition and understanding of such threats difficult.

The solution

Pienso empowers experts to train classification models infused with their nuanced understanding of users and language.

With Pienso, removing hate speech from social and corporate channels is finally a realizable goal.

Monitoring and tracking

Where will the narrative go next?

Research analysts across both government and commercial sectors are often tasked with monitoring complex themes and topics within vast streams of continuous data. As those themes and topics evolve, keeping up becomes even more of a challenge.

Pienso allows fast iteration of models and testing before deployment, meaning your models can evolve as fast as narratives change. And unlike closed-source LLMs, Pienso customers own their models, ensuring their consistency over time.

The solution

For Intelligence Client 2, Pienso made a taskforce of hundreds of expertly-trained LLMs. Each handled a specific aspect of classification, search and summary. 

By using this taskforce to spot patterns in vast streams of global language data, the client  was able to identify  propaganda and track the evolution of narratives designed to sway public opinion.

Prior to Pienso, the team accomplished the analysis manually via collection of spreadsheets. With Pienso, the team has reduced the effort from hours to minutes.

Intelligence Client 2

Governments trust us for good reasons.

The most well-trained LLMs lose value if they can’t be relied on. You need stability and control to track narratives over time. With Pienso, your models are owned by you – which means no updates you haven’t made yourself, and full sovereignty over your data, always.

PromptFactory is your path to GenAI independence.

NLQ and PromptFactory

  • With the launch of PromptFactory, Pienso customers can not only create but also test prompts before deploying them live. This gives them an unprecedented amount of direct control and flexibility, with the ability to adapt models quickly as they need.

  • Through PromptFactory and Pienso’s new NLQ capabilities, anyone can use Natural Language Queries to interrogate their data repositories – making our platform among the first to truly democratize fast and effective GenAI for everyday tasks across industries.

From domain analysts to domain data scientists.

Put the power of GenAI on your side. Pienso helps your analysts and subject experts become model masters and data explorers, whatever their technical proficiency.

  • Explore how people across industries are using Pienso to quickly analyze documents and extract information from vast language datasets.

  • See how companies are using Pienso to find out how their customers are feeling – and improving their customer service (and business success) as a result.

  • Risks are harder than ever to spot. Our clients are keeping up with risks as they evolve, finding ways to ensure their safety and success through Pienso.